> I don't know why you think this is true. If statistically B follows A to a high degree, then a sufficiently advanced statistical model will represent "A then

> I don't know why you think this is true. If statistically B follows A to a high degree, then a sufficiently advanced statistical model will represent "A then B" in some manner. Yes, but suppose A implied B only if C were true. And in the training corpus C were always true (hence learned A=>B) but in the test corpus suppose C is not true, then the learned statistical rule is wrong. The problem is that to cover all these possibilities will require not a billion documents but something like exp(billion) documents (exaggerating, but to give you a feel for combinatorial explosions) as @ClumsyPilot explains. And problem is the real world always keeps “changing” (Eg: whether C is valid or not) compared to the examples you trained on. This is referred to in polite company as “distribution shift” where your underlying statistical distribution keeps evolving.